Flat-Folder finds valid solutions for each connected component of variables via a brute-force search
I wonder if this could get faster by a simple heuristic, for example, by giving some score for each variable and sort the variables to be searched by the score, since this approach worked very well in ORIPA.
In ORIPA, the score is the sum of the number of valid solutions at leaves of subtrees on the search tree, whose root nodes are the variable to give the score. Then as the search progresses, the score will increase. A high score means that the variable assignment is valid for many cases, so if we consume nodes with high scores in early stage, then there will be less failures at deep level of search. This may reduce the number of recursive calls exponentially.
I'm not sure my idea fits to Flat Folder, but I believe there is a chance.
I wonder if this could get faster by a simple heuristic, for example, by giving some score for each variable and sort the variables to be searched by the score, since this approach worked very well in ORIPA.
In ORIPA, the score is the sum of the number of valid solutions at leaves of subtrees on the search tree, whose root nodes are the variable to give the score. Then as the search progresses, the score will increase. A high score means that the variable assignment is valid for many cases, so if we consume nodes with high scores in early stage, then there will be less failures at deep level of search. This may reduce the number of recursive calls exponentially.
I'm not sure my idea fits to Flat Folder, but I believe there is a chance.